refactor: check for metadata before replacing

Prior to replacement, verify the presence of a metadata document in
GBIF. This precaution prevents potential errors when attempting to
replace a metadata document that does not currently exist.

src/gbif_registrar/crawl.py

@@ -3,7 +3,7 @@
 import json
 from requests import post, get, delete
 from gbif_registrar.config import USER_NAME, PASSWORD, GBIF_API, REGISTRY_BASE_URL
-from gbif_registrar.utilities import read_local_dataset_metadata
+from gbif_registrar.utilities import read_local_dataset_metadata, has_metadata
 
 
 def initiate_crawl(local_dataset_id, local_dataset_endpoint, gbif_dataset_uuid):
@@ -53,7 +53,8 @@ def initiate_crawl(local_dataset_id, local_dataset_endpoint, gbif_dataset_uuid):
     # Post a new metadata document to update the GBIF landing page. This is
     # necessary because GBIF doesn't "re-crawl" the local dataset metadata when
     # the new local dataset endpoint is updated.
-    post_new_metadata_document(local_dataset_id, gbif_dataset_uuid)
+    if has_metadata(gbif_dataset_uuid):
+        post_new_metadata_document(local_dataset_id, gbif_dataset_uuid)
 
 
 def post_new_metadata_document(local_dataset_id, gbif_dataset_uuid):

src/gbif_registrar/utilities.py

@@ -1,8 +1,9 @@
 """Miscellaneous utilities"""
 import os.path
+from json import loads
 import pandas as pd
 from requests import get
-from gbif_registrar.config import PASTA_ENVIRONMENT
+from gbif_registrar.config import PASTA_ENVIRONMENT, GBIF_API
 
 
 def initialize_registrations(file_path):
@@ -111,3 +112,27 @@ def read_local_dataset_metadata(local_dataset_id):
     resp = get(metadata_url, timeout=60)
     resp.raise_for_status()
     return resp.text
+
+
+def has_metadata(gbif_dataset_uuid):
+    """Check if a GBIF dataset has a metadata document.
+
+    Parameters
+    ----------
+    gbif_dataset_uuid : str
+        The registration identifier assigned by GBIF to the local dataset.
+
+    Returns
+    -------
+    bool
+        True if the dataset has a metadata document, False otherwise.
+
+    Notes
+    -----
+    The presence of a dataset title indicates that the dataset has been
+    crawled by GBIF and the metadata document has been created.
+    """
+    resp = get(url=GBIF_API + "/" + gbif_dataset_uuid, timeout=60)
+    resp.raise_for_status()
+    details = loads(resp.text)
+    return bool(details.get("title"))

tests/test_utilities.py

@@ -6,6 +6,7 @@
 from gbif_registrar.utilities import initialize_registrations
 from gbif_registrar.utilities import expected_cols
 from gbif_registrar.utilities import read_local_dataset_metadata
+from gbif_registrar.utilities import has_metadata
 
 
 def test_initialize_registrations_writes_to_path(tmp_path):
@@ -52,3 +53,9 @@ def test_read_local_dataset_metadata_returns_str():
     """Test that read_local_dataset_metadata returns a string."""
     metadata = read_local_dataset_metadata("edi.941.3")
     assert isinstance(metadata, str)
+
+
+def test_has_metadata_returns_expected_type():
+    """Test that the has_metadata function returns a boolean."""
+    res = has_metadata("cfb3f6d5-ed7d-4fff-9f1b-f032ed1de485")
+    assert isinstance(res, bool)